Sleep disturbances and depression severity in patients with Parkinson's disease

Abstract Objectives Parkinson's disease (PD) is a multisystem movement disorder associated with sleep disturbance and depression. Sleep disturbances and depression severity share a bidirectional association. This association may be greater in individuals who are more vulnerable to the deleterious consequences of sleep disturbance and depression severity. We investigated whether the association between sleep disturbances and depression severity is greater in patients with PD than in matched controls (MC). Materials and Methods The study sample (N = 98) included 50 patients with idiopathic PD and 48 age‐, race‐, sex‐, and education‐matched controls. Sleep disturbances were assessed using self‐reported total sleep time (TST) on the Pittsburgh Sleep Quality Index, the sleep item on the Beck Depression Inventory, 2nd ed. (BDI‐II), and the Insomnia Severity Index total score. Depression severity was assessed using the BDI‐II total score, excluding the sleep item. Spearman's correlations, chi‐squared tests, and multiple regression were used to assess associations between sleep disturbances and depression severity in PD and MC. Fisher's Z transformation was used to test whether the association between sleep disturbances and depression severity was stronger in patients with PD. Results Shorter TST, sleeping less than usual, and insomnia severity were associated with depression severity in the total sample, r s(94) = −0.35, p = .001; r s(71) = 0.51, p < .001; r s(78) = −0.47, p < .001; r s(98) = 0.46, p < .001, respectively. The association between shorter TST and depression severity was greater in patients with PD than it was in MC, p < .05. Conclusion Short TST may be an important marker, predictor, or consequence of depression severity in patients with Parkinson's disease.

words, these patients may be more vulnerable to the negative impact of sleep disturbances. Preliminary research lends support to this hypothesis. In a Drosophila model of PD, sleep deprivation resulted in greater short-term memory impairments in PD animals compared to controls (Seugnet et al., 2009). Although human research on this hypothesis is sparse, one study found sleep disturbances are both more common and severe among patients with PD than among healthy older adults (Tandberg et al., 1998). These results may suggest patients with PD find sleep disturbances more distressing or impairing than the general adult population. Whether the association between sleep disturbances and depression severity is greater in patients with PD remains unclear. In this study, we explored the associations between different types of selfreported sleep disturbances (differences in TST and insomnia severity) and depression symptoms in a sample of patients with PD and in MC.
We hypothesized that sleep disturbances would be more strongly associated with depression symptoms in patients with PD than in MC.

| Participants
Participants included 50 patients with idiopathic PD and 48 neurologically healthy older adults matched for age, race, sex, and education.
Participants with PD were recruited from the University of Florida (UF) Center for Movement Disorders and Neurorestoration (CMDN) during their routine visits. Clinical diagnosis of idiopathic PD was determined by a movement disorders specialist based on the United Kingdom Parkinson's Disease Society Brain Bank clinical diagnostic criteria using the Unified Parkinson Disease Rating Scale (UPDRS) (Hughes, Daniel, Kilford, & Lees, 1992;Fahn & Elton, 1987). Eligible patients had to demonstrate marked improvement in motor symptoms in response to initiation of dopaminergic therapy, based on the motor subscore of the UPDRS following administration of levodopa. All patients with PD were stabilized on medications for at least 3 months before participation and were in the early to mid-stages of disease severity based on Hoehn-Yahr staging of 3 or less (Hoehn & Yahr, 1967).
Exclusion criteria were (i) inability to provide informed consent; (ii) less than 8th grade education; (iii) evidence of significant neurologic disturbance other than PD (toxic exposure, tumors, encephalitis, epilepsy, stroke in the past year, head injury with loss of consciousness > 24 hr, etc.); (iv) other clinically significant medical problems such as metastatic cancer, human immunodeficiency virus, blindness/deafness, liver or renal disease, severe anemia, myocardial infarction (< 6 months), congestive heart failure (functional stage > 3), and so forth; (v) severe psychiatric disease including significant history of substance abuse or current history of untreated alcohol or substance abuse, severe major depressive disorder determined by a cutoff score > 19 on the Patient Health Questionnaire-9 (Kroenke, Spitzer, & Williams, 2001), bipolar disorder, schizophrenia, or posttraumatic stress disorder; (vi) reduced cognitive status based on scores less than 25 on the Mini-Mental State Exam (MMSE); (vii) inability to read and understand English; (viii) brain surgery (deep brain stimulation surgery or pallidotomy). Additional exclusion criteria related to sleep included (i) self or bed partner report of symptoms on the Pittsburgh Sleep Quality Index (PSQI), diagnosis, or treatment for obstructive sleep apnea (OSA), restless leg syndrome (RLS), or periodic limb movement disorder (PLMD) or (ii) lack of bed partner who was able to document the absence of OSA, RLS, and PLMD.
There were no exclusions for medication use.
Disease-specific information about the PD participants was obtained from an IRB-approved clinical research database maintained by the UF CMDN. Obtained information included standard clinical measures for staging the severity of motor symptoms and disease course:
Control participants were recruited through interested patient spouses, newspaper advertisements, community fliers, local aging communities, and mailers. Individuals were excluded for the above exclusion criteria of the PD group or if they had a history of a movement disorder. Table 1 shows the demographic characteristics of the PD and MC groups. The patients with PD were predominantly men (76%), 50 to

| Total sleep time
Item number 4 of the PSQI was used to assess average TST participants typically obtained per night over the past month. This item has been used in numerous studies as a measure of subjective TST (e.g., Nebes et al., 2009). Total sleep time, in minutes, was used as an independent variable in this study. 1988). Responses to 16a (sleeping more than usual) or 16b (sleeping less than usual) >0 were recorded as indicating sleeping more or sleeping less than usual. This item was used to derive two variables:

| Sleep changes
(i) sleeping more than usual or no change and (ii) sleeping less than usual or no change. Thus, for each of these variables, the change in sleep was compared to those who reported no change in sleep.

| Insomnia symptom severity
The Insomnia Severity Index (ISI) is a seven-item questionnaire used to assess sleep quality and insomnia severity over the previous 2 weeks (Bastien, Vallieres, & Morin, 2001). The ISI is a continuous measure with scores ranging between 0 and 28. The independent variable of interest was total ISI score.

| Depression symptom severity
The BDI-II, a well-validated 21-item self-report inventory, was used in this study to quantify the severity of depression symptoms. The BDI-II is widely used to assess depression in older adult samples and a Movement Disorders Society task force recommended the BDI as a valid continuous measure of depression severity in PD (Schrag et al., 2007). An adjusted total BDI-II score, one that did not include item 16 (i.e., changes in sleep), served as the dependent variable in this study.

| Analyses
IBM SPSS 24 statistical software was used to analyze these data.
One-way analyses of covariance (ANOVA) and chi-squared tests were used to assess group differences in demographic and health features when variables were normally distributed or categorical, respectively. Sleep variables, MMSE, BDI-II, average number of medical conditions, and average number of medications were not normally distributed within each group (Kolmogorov-Smirnov < .05); Mann-Whitney U test was used to assess group differences in these variables (Table 1). In the total sample, Spearman's correlation coefficient was used to determine whether different types of sleep disturbances were associated with each other, depression symptom severity, or sex. Spearman's correlation coefficient was also used to determine whether sleeping more or less than usual, compared to no changes, was associated with depression symptoms in each group. Because previous research found an association between long TST (≥ 9 hr) with depression in PD and because long TST was rare in our sample, the four individuals with TST ≥ 9 hr were excluded from analyses used to investigate the association between TST and depression symptoms severity. To test whether the strength of the correlation between sleep disturbances (TST or ISI) and depression severity were different in patients with PD than MC, we conducted Fisher's Z transformation for Spearman's correlation coefficient (Snedecor & Cochran, 1980). Sensitivity analyses, with these extreme cases excluded, were conducted to test whether outliers influenced the results. We also conducted sensitivity analyses excluding individuals using antidepressants or sleep aids.
A follow-up moderation analysis using the PROCESS plugin for SPSS (Model 1) was conducted to test for main effects of ISI and TST and their interactions on depression severity: BDI-II adjusted total score was the dependent variable; TST and ISI were the independent variables.    Moderation analysis within the PD group found no significant interaction between ISI and TST in predicting depression severity.

| RE SULTS
With both ISI and TST included in the model, only shorter TST was a significant predictor of depression severity in PD.

| D ISCUSS I ON
In this study, we investigated whether the association between sleep disturbance and depression severity is greater in patients with PD than in MC; we hypothesized that the association would be stronger in patients with PD. We found that shorter TST was associated with greater depression severity in patients with PD than in MC. The rates and severity of sleep disturbance in our PD sample were comparable to previous reports using the PSQI, with over half (58%) reporting sleep disturbance (Borek et al., 2006;Ferreira et al., 2006;Ratti et al., 2015). Consistent with previous studies, patients with PD had more severe insomnia, higher rates of sleeping more than usual, and more symptoms of depression than MC (Ferreira et al., 2006;Tandberg et al., 1998). The absence of a group (PD vs. MC) difference in TST is consistent with previous reports (Happe et al., 2001;Ratti et al., 2015). There is a compelling bivariate association between insomnia and depression severity in PD (Chung et al., 2013;Rutten et al., 2017;Suzuki et al., 2009). Although insomnia and TST were associated with each other and with depression severity, insomnia severity was not more strongly associated with depression in PD than it was in MC in this present study. When TST and ISI were considered together, it was shorter TST that accounted for the majority of the variance in depression severity.
One possible mechanism through which shorter TST may be associated so strongly with depression severity in PD relates to dopamine dysfunction. Interestingly, shorter TST was not associated with sleeping less than usual reported in patients with PD, suggesting the greater association between short TST and depression in PD may be more associated with persistent sleep loss rather than recent changes. In normal individuals, dopamine levels increase in the brain following sleep loss, a putative compensatory mechanism (Drummond et al., 1999;Volkow et al., 2009). Dopaminergic dysfunction associated with PD pathology may reduce these patients' ability to compensate for the effects of sleep loss, thereby increasing risk for greater depression severity. The cross-sectional design of this study prevents us from establishing causal relationships or establish why shorter TST is more strongly associated with depression severity in PD than MC. Nevertheless, the results of this study highlight the importance of considering TST as a potential predictor, marker, or consequence of depression severity in these patients.
These findings must be interpreted in the context of several with self-report, we cannot rule out the possibility that a common factor of self-report was involved in these findings. However, such an explanation cannot explain the unique association between shorter TST and depression severity in PD compared to MC, an association that was not observed for insomnia severity and depression severity in PD.

ACK N OWLED G M ENTS
We thank Jonathan Trout for data management and editorial comments on this manuscript.